Does life in the U.S. take a toll on health? Duration of residence and birthweight among six
decades of immigrants
Julien Teitler
Melissa L. Martinson
Nancy E. Reichman
DRAFT
DO NOT QUOTE OR CITE WITHOUT PERMISSION
This research was funded by NICHD Grant #R03HD058067 and received additional support
from the Columbia Population Research Center (#R24HD058486), the Center for Health and
Wellbeing at Princeton University, and the Office of Population Research at Princeton
University (#T32HD001763). The authors are grateful to Kevin Lin for excellent research
assistance.
1
Abstract—
We use the 1998-2009 National Health Interview Survey to systematically investigate
cohort differences in low birthweight among U.S.-born children of mothers arriving in the U.S.
between 1955-2009 and giving birth between 1980-2009, cohort-adjusted patterns in low
birthweight by maternal duration of residence in the U.S., and cohort-adjusted patterns in low
birthweight by maternal duration of residence stratified by age at arrival and region of origin.
These analyses are necessary for understanding immigrant women’s health trajectories, as
proxied by their infants’ birthweight, and for understanding how living in the U.S. affects health
more generally. We found a consistent deterioration in infant health with successive immigrant
cohorts, a curvilinear association between duration and low birthweight controlling for cohort
effects, that Asians benefit more than Hispanics from migrating to the U.S., and that low
birthweight decreases with duration among women who came to the U.S. as adults but increases
with duration among those who came as children. The findings underscore the need to
disaggregate immigrant groups when studying post-migration trajectories and provide additional
evidence that early life exposures are a key to understanding why the U.S. lags other developed
nations in health.
2
Introduction
Despite having the highest per capita healthcare expenditures in the world, the U.S. ranks
poorly on most health indicators compared to other developed countries. The U.S. disadvantage
spans the lifecourse, from infant mortality to life expectancy at older ages (United Nations 2011),
and is not limited to mortality. Differences in morbidity are also apparent across the lifecourse,
from very young ages through late adulthood, at least compared to England (Martinson, Teitler
and Reichman 2011). The most obvious potential explanations— namely, smoking, stress,
inequality, health insurance and health care, violence, and diet—have not gone far in explaining
the persistent and generalized U.S. health disadvantages (National Research Council 2010;
Banks et al. 2006; Martinson, Teitler and Reichman 2011), leading to speculation that there is
something generally intrinsic or “toxic” about living in the United States.
Immigrants provide a valuable lens through which to learn about how living in the United
States affects residents’ health. Research on immigrant health suggests that migrating to the U.S.,
even from low-income countries with poor quality healthcare systems, takes a toll on one’s
health. When they land on U.S. soil, the health of immigrants is better than that of U.S.-born
individuals of the same race/ethnicity. However, the immigrant health advantage appears to
erode over time (e.g., Antecol and Bedard 2006; Cho et al. 2004; Day et al. 2006; Goel et al.
2004; Uretsky and Mathiesen 2007). Because immigrants arrive at different ages, come from
different countries, are of different ethnicities, come with different levels of human capital, and
have different durations of residence, their experiences can help us understand how living in the
United States affects health. That is, the considerable heterogeneity of the immigrant population
provides a useful lever for learning about “for whom” and “under what circumstances” exposure
3
to the U.S. has detrimental effects on health—questions that cannot be easily answered from
studies of U.S. natives or from cross-national comparisons.
Despite considerable interest in immigrant health and the production of a large body of
theoretical and empirical research over recent decades, credible estimates of the effects of
duration of residence in the U.S. on immigrant health have yet to be produced. The existing
literature has, almost without exception, failed to account for differences in health endowments
of immigrants arriving in different time periods (cohort effects), which could seriously bias
estimated duration effects.
In this paper, we investigate how a highly salient measure of immigrant health – low
birthweight of children born in the U.S. to immigrant mothers – has varied across arrival cohorts,
is associated with duration of residence adjusting for cohort differences, and varies by region of
origin and age at arrival. All three analyses are critical for moving the research on immigrant
health trajectories forward. Additionally, the region-of-origin and age-at-arrival analyses exploit
important variations in immigrant experiences to help us learn about U.S. health disadvantages
more generally.
We focus on birthweight for substantive and practical reasons. The substantive reasons
are that: (1) Native/immigrant differences in birth outcomes mirror those in adult health;
specifically, birthweight-related outcomes of foreign-born mothers are uniformly more favorable
than those of native-born mothers of the same race or ethnicity (e.g., Cho and Hummer 2001;
David and Collins 1997; Landale, Oropesa and Gorman 1999; Landale, Oropesa, Llanes and
Gorman 1999; Singh and Yu 1996). (2) Some of the key behaviors thought to be affected by
immigrant acculturation (e.g. diet and smoking) are important predictors of birthweight
(Reichman 2005). (3) The maternal-child link is potentially important for understanding patterns
4
in immigrant health across generations. (4) The share of children in the U.S. born to immigrant
mothers is large, with one quarter of births in the U.S. in 2008 taking place to women born
outside of the 50 U.S. states or DC (Martin et al. 2010). The practical reasons are that: (1)
Birthweight is a widely-used and readily-available index of infant health that is strongly
associated with subsequent morbidity (Reichman 2005). (2) Maternal reports of birthweight are
highly accurate, even retrospectively after many years (McCormick and Brooks-Gunn 1999;
O’Sullivan, Pearce and Parker 2000). (3) Because retrospective reports are reliable, information
obtained from mothers about the birthweight of their children going back to when those children
were born can be used, in conjunction with the mothers’ years of arrival, to construct long
duration time series using available nationally-representative data.
Background
Immigrant Health Trajectories
As indicated earlier, most existing studies of immigrant health trajectories have found
negative associations between duration of residence and health; specifically, the longer
individuals live in the U.S., the worse their health. There is some evidence of similar associations
in Canada (McDonald & Kennedy 2004; Newbold 2005) and Australia (Chiswick, Lee and
Miller 2008; Julian and Easthope 1996). However, Jasso et al. (2004) found evidence of a
curvilinear association, suggesting improvements in health soon after arrival to the U.S. followed
by steady deterioration after about five years. A handful of studies have investigated associations
between mothers’ duration of residence and the birth outcomes of their U.S.-born infants. Most
have found evidence of worse outcomes with longer duration (e.g., Ceballos and Palloni 2010;
Landale, Oropesa and Gorman 2000), although a recent article based on data from three national
birth cohort studies and using more refined duration intervals than the other studies found
5
curvilinear patterns in birthweight akin to those found by Jasso for adult health (Teitler, Hutto
and Reichman 2012).
The dominant explanation for the purported decline in immigrant health by duration is
acculturation, which has been characterized in much of the empirical literature using measures of
language acquisition, generational status, and age of arrival, all of which are highly associated
with duration of residence in the host country. While acculturation theory was much more
nuanced in its original theoretical formulation (see Lara et al. 2005; Portes and Rumbaut 2001),
the empirical literature often distills the concept into one of gradual acceptance of the new
culture as the protective effects of the country of origin dissipate over time. That is, the process
is assumed to be monotonic and assimilative in nature. Acculturation is thought to operate
primarily through individuals’ health behaviors, particularly drug use, alcohol abuse, cigarette
smoking, and unhealthful dietary patterns (see Lara et al. 2005). Specifically, the longer
immigrants reside in the U.S., the more likely they are to engage in those behaviors, which in
turn compromise their health.
There is evidence that immigrants’ age at arrival is associated with health-related
outcomes, including cognitive development (Glick et al. 2009) and obesity (Kaushal 2009;
Roshania et al. 2008), as well as outcomes that may affect health, such as language proficiency
(Bleakley and Chin 2010). There is also strong evidence that decisions about whether to engage
in health-related behaviors (including those related to birth outcomes, such as smoking) are made
during adolescence (Elders et al. 1994; Kandel et al. 1998). The critical period for take-up of
smoking appears to be narrow and a fairly universal phenomenon (World Bank 1999). If healthrelated behaviors are driving associations between duration in the U.S. and health, then we
6
would expect immigrants who came to the U.S. during or prior to adolescence to be most
negatively affected by duration in the U.S.
Changes in behaviors are not the only processes that can potentially shape immigrants’
health trajectories. Exposures to discrimination or other potentially health-compromising social
stressors could also take a toll on immigrants’ health (Nazroo 2001; Singh and Siahpush 2002;
Reijneveld 1998; Uretsky and Mathiesen, 2007). However, health could also be favorably
affected by migrating to the U.S. if the move leads to economic opportunities and better access to
high-quality healthcare. Thus, factors such as race/ethnicity, education, and circumstances
surrounding the decision to migrate, all which are associated with place of origin, may moderate
duration effects on health.
Cohort effects
All of the above-mentioned studies of immigrant health trajectories are based on crosssectional data, leaving open the possibility that observed associations between duration in the
U.S. and health could be driven (or partly suppressed) by patterns in health selectivity of
immigrants arriving to the U.S. over time, since duration in the U.S. is a function of year of
arrival. For example, if the health of new entrants to the U.S. declines over successive
immigration cohorts and this pattern is not accounted for, estimated duration effects of U.S.
residence would be positively biased (i.e., increased duration would appear to improve health).
Conversely, if the health of new entrants to the U.S. improves over time, estimated duration
effects that do not account for this trend would be negatively biased. To properly estimate the
effects of duration of residence in the U.S. on health, it is necessary to account for year-of-arrival
effects.
7
The extent to which the health of immigrants varies by cohort represents a significant
knowledge gap, as does the extent to which differences across cohorts explain associations
between immigrant duration of residence and health. Jasso et al. (2004) found that self-rated
health among immigrants who came to the U.S. in 1991 was better than that among immigrants
who arrived in 1996, and Antecol and Bedard (2006) found some evidence of worsening health,
particularly among women, across four recent cohorts of immigrants. Hamilton and Hummer
(2011) found no clear pattern across immigration cohorts for blacks. These findings suggest that,
overall, immigrants have become less positively selected on the basis of health over time;
however, they are based on too few data points and from observation periods too short to allow
for general inferences to be made. For this reason, the findings from these studies cannot be used
to ascertain the extent to which cohort effects bias estimates of immigrant health trajectories.
Disentangling duration and cohort effects on immigrant health requires repeated cross-sectional
data over a large period of time with granular duration and year-of-arrival intervals.
Age effects
Age is an important potential source of confounding when estimating the effects of
duration, as the duration and age are perfectly collinear. The association between maternal age
and low birthweight has a distinct “U” shape, with high rates among teenagers and older mothers
and low rates among women in their 20s and early 30s (Reichman 2005; Reichman and Pagnini
1997). Thus, age is likely to confound estimated effects of duration of U.S. residence on low
birthweight, unless age is controlled for or observations are confined to the middle “healthy” age
range.
Period effects
8
It is important to consider whether observed differences in health (in our case, low
birthweight) across immigrant cohorts reflect secular changes in the U.S. as a whole. If that is the
case, observed cohort differences could reflect a more general and less immigrant-specific trend.
As can be seen in Figure 1, which documents rates of low birthweight in the U.S. from 1950–
2009 (compiled from U.S. natality data), the trend has remained quite flat. The rates have
fluctuated only within a narrow corridor of about 7–8 percent over 60 years, despite many
changes over the period that would be expected to affect rates of low birthweight over time, such
as: (1) tremendous changes in cigarette smoking, which increased until around 1960 and halved
over the following 40 years (Burns et al. 1997); (2) increased use of assisted reproductive
technology, which has increased the rate of multiple birth (a risk factor for low birthweight)
(Martin, Hamilton and Osterman 2012); (3) increased rates of intervention and delivery of
infants prior to term in cases of poor fetal growth (Joseph, Demissie and Kramer 2002); and (4)
substantial changes in the sociodemographic composition of the population as a whole and of
births in particular (Congressional Research Service 2011; Livingston and Cohn 2010). The
remarkable stability of the trend in low birthweight despite these major forces suggests that
period effects are unlikely to bias our estimated cohort and duration effects on low birthweight.
Our contribution
In this paper, we systematically investigate (1) cohort differences in low birthweight
among children born to immigrant mothers arriving in the U.S. between 1955 and 2009, using
fine-grained duration and year-of-arrival intervals; (2) patterns in low birthweight by maternal
duration of residence in the U.S., controlling for cohort differences; and (3) patterns in low
birthweight by maternal duration of residence stratified by age at arrival and region of origin,
also controlling for cohort differences. All of these analyses are necessary for ascertaining the
9
extent to which life in the U.S. takes a toll on immigrant women’s health, as measured by their
infants’ birthweight. The analyses stratified by age at arrival and place of origin also represent
important steps for determining how living in the United States affects health.
Data
We use the 1998 to 2009 waves of the National Health Interview Survey (NHIS) to
investigate the relationships between maternal duration of residence in the United States, year of
arrival, and low birthweight. The NHIS is the largest household health survey in the U.S. and is
widely used to monitor national health trends. The restricted version of the NHIS is used for the
analyses in this study because it includes the actual year of arrival of the mother, in contrast to
the public use version which includes this variable only in 5 year intervals and is top-coded at 15
years or more in the U.S. Using all available years of the restricted NHIS that include the
requisite analysis variables provides us with a unique and invaluable data source for
disentangling effects of duration of residence in the U.S. from those of year of arrival. Because
1998 was the first year that the NHIS asked about year of arrival and the year of birth was
available only for (randomly-selected) children under age 18 residing with their mothers at the
time of their interviews, our analysis sample of 21,700 mothers consists of women who came to
the U.S. between 1955 and 2009 and gave birth in the U.S. between 1980 and 2009. Table 1
shows sample sizes by 5-year arrival intervals, rounded to the nearest 100 to conform to
institutional protocols governing the presentation of sensitive data. The NHIS uses a multistage
probability sample; consequently, all figures and analyses have been weighted accordingly using
the svy commands in Stata SE 11.
We focus on low birthweight (< 2500 grams, or about 5.5 pounds) as the outcome. Low
birthweight is a widely used, well-measured, and reliably-reported marker of poor health at birth.
10
It is the second leading cause of infant mortality in the U.S. after birth defects and is associated
with long-term health and developmental problems among infants who survive (Reichman,
2005). As indicated earlier, maternal retrospective reports of their children’s birthweight—which
we use for our analyses—are known to be reliable, even after many years.
Most of the results presented in this paper are based on the mother’s actual year of arrival
collapsed into 5-year-arrival bands in order to comply with the NHIS and institutional
restrictions regulating the presentation of data based on small cell sizes. However, analyses
conducted using narrower intervals (and even individual years) produced results highly
consistent with those using the 5-year intervals.
Duration of residence in the U.S. is calculated by subtracting the mother’s year of arrival
in the U.S. from the year of the focal child’s birth. Births that took place before the mother came
to the U.S. are not included in the sample. Because only the years (and not months) of arrival and
birth are available, our duration measures, which are expressed in single (whole number) years,
actually encompass 24-month duration periods. For example, women who gave birth 12 months
plus 1 day after arriving, those who gave birth 36 months less one day after arriving, and those
with arrival-to-birth intervals within that range would all be coded as having lived in the U.S. for
2 years at the time of the birth (that is, having a two year difference between arrival and the
birth). In our graphical analyses, following Teitler, Hutto and Reichman (2012), we use the
following duration categories: 0–2, 3–5, 6–10, 11–15, and 16+ years. In our regression analyses,
we include single year duration measures, omitting the 1-year category (which represents the
most recent arrivers, who had been in the U.S. 0 to 24 months when they gave birth to the focal
child1).
1
Since the calculation of duration of residence in the U.S. is based on calendar year of immigration and calendar
year of birth, the range of exposure to the U.S. captured by each of our 1-year duration of residence variables is
11
Analyses
The first two questions we address in this paper are: (1) To what extent has low
birthweight of immigrants changed across arrival cohorts over the past 50 years? (2) To what
extent does low birthweight vary by duration of residence in the U.S.? To answer the first
question, we compute rates of low birthweight for 5-year arrival cohorts within specific duration
intervals. To answer the second question, we estimate cohort-adjusted duration effects using
logistic regression to estimate low birthweight as a function of both the number of years the
mother had been in the U.S. at the time of the birth and her year-of-arrival cohort (as 5-year
interval dummy variables and, in alternative specifications, as a continuous variable). We also
control for maternal age (less than 20, 20–34, or 35+ years). We then graphically display the
fitted 2nd order polynomial curve corresponding to the logit coefficients.
In our third set of analyses, we stratify by age at arrival and by region. Specifically, we
conduct separate analyses for mothers who arrived in the U.S. as minors (< 18 years old) and as
adults (over age 18), and for two region-of-origin groups for whom sample sizes are sufficient:
(1) mothers from Mexico, Central, South America, and the Caribbean, and (2) mothers from
Asia, Southeast Asia, and the Indian subcontinent. For ease of exposition, we refer to the former
as “Hispanics” and the latter as “Asians,” recognizing that these terms do not accurately
characterize all individuals from those areas. Finally, although sometimes considered foreignborn in research on immigrant health, island-born Puerto Ricans, who are U.S. residents, are not
included in the Hispanic group.
Results
greater than one year. To illustrate, a birth in 1999 to a mother who arrived in 1998 would be coded as having
resided in the U.S. for one year. This could be a mother who arrived on December 31st 1998 and gave birth on
January 1st 1999 (so arrived in the U.S. 1 day before giving birth), but could also be a mother who arrived January
1st 1998 and gave birth on December 31st, 1999 (so in the U.S. 2 years before giving birth). Thus, each 1-year
duration of residence variable represents the median potential exposure to the U.S. within a 24 month period.
12
Cohort effects
Both official statistics and the survey data from the NHIS demonstrate that there have
been important changes over time in the characteristics of immigrants arriving to the United
States. Figures 2 and 3 present data from various U.S. government reports (see table notes) on
the composition of legal immigrants to the U.S., by world region, from 1950 to 2010. With the
exception of a peak in the number of legal immigrants from Mexico between 1988 and 1992,
coinciding with the Legally Authorized Worker and Specialized Agricultural Worker amnesty
programs under the Road to the Immigration Reform and Control Act, the proportions of
immigrants to the U.S. from Asia, all Spanish speaking countries, and Europe have remained
fairly stable since the 1970s (Figure 2). The proportion has increased notably only for
immigrants from Africa, but this group remains small in comparison to other sending regions.
Most of the major regional compositional changes of new immigrants occurred from 1950 to
1975, during which there were large increases in the numbers from Spanish speaking and Asian
countries (Figure 3). The absolute numbers of immigrants from Europe and Canada have
remained relatively stable since the 1950s, but as the total number of immigrants has risen, their
proportional representation has diminished substantially.
Data on country or region of origin in our NHIS analysis sample (immigrant women
giving birth in the U.S. between 1980 and 2009) confirm the relative stability of proportions
from various sending areas. However, the NHIS data (which may include undocumented
immigrants) suggest that the share from Mexico, Central America, and South America has been
substantially larger among immigrant women giving birth in the U.S. than among legal
immigrants to the U.S. as a whole—60–65 percent (Figure 4) compared to approximately 40
percent per year (Figure 3).
13
The relative stability in the regional composition of immigrants over the recent past does
not mean that the composition of immigrants has otherwise remained the same. We find in our
NHIS sample that cohorts have become much less educated over time; 94 percent of immigrant
mothers in the 1955–1960 arrival cohort had a high school education (assessed at the time of the
NHIS interview), compared to 57 percent of those in the 2005–2009 cohort (Figure 5). Since
education and health are strongly intertwined, the decreasing level of education among
immigrants—if not accounted for—is likely to confound estimated effects of duration of
residence in the U.S. on birth outcomes. The same could be true for any other compositional
changes among immigrants that are related to health. Education is particularly difficult to
account for in studies of immigrant health trajectories that do not focus on one particular sending
country because educational systems vary across sending countries and equivalencies are
difficult to assess, and because reporting on educational attainment in surveys with U.S.-based
educational response choices may vary by region of origin. Additionally, and more specific to
the NHIS and our study, some mothers may have arrived prior to having completed their
education. The most direct and expedient way to account for all observed and unobserved
compositional differences when investigating the effects of duration of residence on health is to
control for cohort, which is the approach we follow.
In Figure 6, we present rates of low birthweight among U.S.-born children of immigrants
over successive immigration cohorts. We focus on mothers in specific duration intervals to purge
duration effects from actual cohort differences. What is notable from this figure is that there has
been a consistent deterioration in infant health with successive immigrant cohorts. That is, rates
of low birthweight steadily increased over the 50-year observation period. For example, among
women giving birth 1–2 years after arrival to the U.S., the rate of low birthweight was 9.6
14
percent for the 2005–2009 cohort, compared to 7.6 percent for the 1985–1990 cohort. A similar
pattern is apparent within almost all of the other duration categories. The exception is for
mothers giving birth 16 years or more after coming to the U.S., for whom rates of low
birthweight were substantially lower among the last cohort (1985–1990) than among the
previous cohorts. However, that “reversal” is likely to be an artifact of women in the last cohort
being constrained to the beginning of the 16+ duration interval whereas those in earlier arrival
cohorts were not. To the extent that there are detrimental duration effects among long-term
residents (which is not inconsistent with what we find later in this paper), the censored
composition of this group would suppress rates of low birthweight. Relative risk analyses (not
shown) confirm that year-of-arrival effects are large and highly significant. Specifically, we find
that the risk of delivering a low birthweight infant increased by 1.1 percent per year, controlling
for duration of U.S. residence in single year increments. At the aggregate level, the percentage
point increase in low birthweight, controlling for duration, was .10. That is, children of
immigrants who were low birthweight increased by approximately 1 percentage point every 10
years during the latter half of the 20th century and first decade of the 21st century.
Effects of duration in U.S.
To estimate the effects of duration of residence in the U.S. on low birthweight, we
estimate logistic regressions of low birthweight on 14 single-year duration indicator variables
plus an indicator for living in the U.S. 16 years or more, with less than 1 year in the U.S. as the
reference category. We control for the eleven 5-year arrival cohorts in Table 1 using indicator
variables for 10 of those cohorts and the other serving as the reference category, and for maternal
ages at the time of the birth of less than 20 and greater than 34 years, with 20–34 years as the
reference category. In addition to the model for the overall sample, we estimate corresponding
15
models that alternatively restrict the sample to Hispanic mothers (defined earlier as being from
Mexico, Central and South America, or the Caribbean); Asian mothers (defined earlier as being
from China, Southeast Asia, or the Indian subcontinent); mothers that arrived prior to age 18;
and mothers that arrived at age 18 or older. We then estimate and plot a quadratic regression
curve of the logistic regression coefficients of the duration indicator variables to visually display
the pattern of associations between low birthweight and duration of residence in the U.S.
controlling for cohort effects and maternal age. The graphical results are presented in Figures
7a–e and the full regression results upon which the five figures are based are included in
Appendix Table 1.
Our full-sample analysis reveals a pattern similar to that found by Teitler, Hutto and
Reichman (2012) based on three national datasets (ECLS-B, ECLS-K, and Fragile Families), but
not accounting for cohort effects. Specifically, there appears to be a curvilinear association
between duration in the U.S. and low birthweight (Figure 7a). The logged odds that immigrant
women give birth to low a birthweight infant decreases over the first 11 years in the U.S. and
subsequently increases. The odds of having a low birthweight baby 11 years after arriving are
almost half (e-.49 = .61) the odds for women giving birth less than 2 years after arrival. Even after
being in the U.S. 16 or more years, the rate of low birthweight, conditional on maternal age and
arrival cohort, is no worse that it was than upon arrival (logit coefficient = -.13). Every duration
coefficient (i.e., the log odds of low birthweight for a given duration relative to being in the U.S.
less than 2 years) is negative, and most are statistically significant at p < .05 or close to
significant. As an alternative specification, we restricted the sample to women ages 20–34 – the
range within which age effects on low birthweight are very small – rather than controlling for
maternal age and the results were substantively unchanged (not shown).
16
Figures 7b and 7c (and the corresponding models in Appendix Table 1) show the duration
coefficients for Hispanic- and Asian- origin mothers, respectively. While the regression-fitted
curvilinear trend is less pronounced for these subgroups than for the full sample, it still appears
that there are beneficial effects of duration over the first 9–11 years which subsequently taper
off. As for the full sample, the rates of low birthweight after 15 years are no greater than they
were upon arrival.
An important and striking difference revealed by the stratified analyses is that the
apparent health benefits accruing from additional years in the U.S. are particularly large for
Asian immigrants, for whom the logit coefficients for duration are approximately 4 times larger
than those for Hispanics. This is a particularly noteworthy finding as most studies on immigrant
health have focused on Hispanics (often Mexicans), possibly providing a negatively-skewed
view of immigrant health trajectories in the U.S. That said, stratifying by region reduces the
analysis sample sizes significantly, and as a consequence, the duration estimates are not
statistically significant in the model for Hispanics. For Asians, many of the duration coefficients
are statistically significant as a result of their relatively high magnitudes. Finally, it is important
to point out that the much larger coefficients for Asians reflect a large decline in rates of low
birthweight between the first and second years in the U.S. Were we to use a combined 1–2 year
duration interval as the reference category, the difference in the magnitude of coefficients
between the two regional groups would be largely erased.
The last two figures (7d and 7e) show results from models stratified by maternal age at
arrival. As discussed earlier, there are theoretical reasons to expect that duration effects would be
different for immigrants who come to the U.S. before or during the particularly impressionable
adolescent years than for those who come after they have already reached adulthood and may
17
have already established lifelong health-related habits. It is also possible that selectivity on the
basis of health is very different for adults (who make the decision to migrate) than children (who
passively accompany their parents or other adults) and that the effect of exposure to the U.S.
depends on the individual’s health endowment.
The results presented in Figures 7d and 7e (and in the corresponding models in Appendix
Table 1) demonstrate that the health trajectories are, in fact, distinctly different for the two ageat-arrival groups in a manner that is consistent with acculturation theory and with what is known
about critical periods for the formation of health behaviors. While the likelihood of having a low
birthweight infant decreases with duration among women who came to the U.S. as adults, the
likelihood of low birthweight increases with duration among those who came to the U.S. as
children (all but one coefficient is positive though most are not statistically significant due to the
greatly reduced sample size). Immigrants coming to the U.S. as children account for most of the
long-exposure sample (i.e., women having births more than 10 years after arriving) and those
arriving as adults account for most of the short-exposure sample (i.e., women having births fewer
than 10 years after arriving). The fact that the associations are divergent for the two groups
explains the U-shaped association between duration and low birthweight observed in the full
sample. That is, the initial decline in low birthweight reflects the experience of women who
arrived in the U.S. during their prime reproductive years and gave birth soon afterward, while the
increase after 10 years in the U.S. reflects the experiences of those who arrived at much younger
ages.
As an alternative specification, we re-estimated all of the regression analyses using a
continuous measure of year of arrival (rather than 5-year cohorts) and the results were
18
substantively unchanged, suggesting that—at least over the past 50 years—the declining levels
of immigrants’ human capital can be thought of as a gradual and persistent phenomenon.
Discussion
A fair amount of scholarly work over recent decades has focused on documenting and
attempting to explain health declines of immigrants to the U.S., within and across generations.
This research has been based on the notion that the acculturation process erodes immigrants’
initial health advantages over time. The literature to date can be characterized as having three
fundamental limitations – not having taken into account variations in health across arrival
cohorts, using very broad duration intervals (which is what are available in most relevant
datasets), and failing to account for important interactions. Accounting for cohort effects is
critical because year of arrival is a linear function of duration of residence and because there
have been profound changes in the composition of immigrants over time. We found a
pronounced and consistent deterioration in infant health with successive immigrant cohorts
between 1955 and 2009. Using broad and fixed duration intervals can conceal complex nonlinear associations between duration and health. We found that the use of fine-grained duration
intervals (even single years) paints a nuanced picture of the associations between duration and
low birthweight. Not allowing for associations to vary across groups can mask substantial
heterogeneity of effects within the immigrant population. In addressing all three of these
limitations, we found evidence of a nonlinear association between duration of residence in the
U.S. and low birthweight that appears to reflect two very different processes—improvements in
health with duration among immigrants who came to the U.S. as adults and deteriorations in
health with duration among those who arrived as children. We also found differential duration
effects by place of origin for the two largest broad groups of immigrants in the U.S. In particular,
19
Asian immigrants appear to benefit more than Hispanic immigrants, in terms of health, after
arriving to the U.S.
The divergent results for the two age-at-arrival groups allow us to pinpoint a clearly
identifiable group for which health erosion with duration in the United States takes place—those
who came to the U.S. as minors (sometimes referred to as generation 1.5). These findings are
consistent with any the following scenarios: Adults making decisions to come to the U.S. likely
do so with the expectation of improving their circumstances and a strong motivation to take
advantage of new opportunities. They are primed to benefit from increased opportunities and
resources, and at the same time, relatively protected from commercial and social pressures to
engage in health-compromising behaviors (e.g., smoking, substance use, poor diet, sedentary
lifestyle), having already established their health habits. Immigrant youth, who come to the U.S.
with their parents or other adults, are less likely to be as highly motivated to benefit from the
migration experience. They also arrive at a stage in the lifecourse at which they are more likely
to succumb to influences to engage in health-compromising behaviors. Finally, it is possible that
children are particularly vulnerable to health compromising social and physical environments
that may exist in the United States.
The finding that Asians experience larger gains to being in the U.S. than Hispanics is
noteworthy and novel, as Asian immigrants are an understudied group despite their growing
numerical representation in the U.S. Among legal immigrants, Asians now outnumber Hispanics.
The bulk of research on immigrant health to date has focused on Hispanics. Our finding of
region-of-origin differences in duration effects underscores the need for more studies of Asian
immigrant health trajectories and the underlying processes.
20
This study focused on one indicator of health, infant birthweight, which represents a
substantively important outcome and allowed us to create a nationally-representative dataset with
substantial variation in year of arrival, duration of U.S. residence, maternal age at birth, age at
arrival, and region of origin—all of which were necessary for accomplishing the aims of this
study. That said, it is important to point out that the results based on this outcome cannot
necessarily be generalized to other aspects of physical or mental health. It is possible that
duration of residence in the U.S. is associated with benefits to reproductive health but with
decrements to other measures of health, although there is no reason to expect that would be the
case. It is also possible that the effects of duration are different for men than for women,
particularly since women may be more connected to healthcare services in their reproductive
years. Finally, in terms of birthweight itself, it is not possible to explore the specific etiologies
(preterm birth, intrauterine growth retardation) behind the observed duration effects with
available nationally-representative data.
In addition to addressing questions about our primary focus—immigrant health—the
findings from this study have implications for the vexing observation that the health of
Americans compares unfavorably to that of residents of most other developed countries. The fact
that the health of immigrants who come as adults does not appear to deteriorate by virtue of
being in the U.S., but that the health of those who come as children does seem to erode over
time, suggests that child immigrants are less favorably selected on health and therefore less
resilient to health compromising exposures or that U.S. health disadvantages are embedded early
in the life course. The latter is consistent with recent findings by Martinson, Teitler and
Reichman (2011) that U.S. health disadvantages (at least compared to England) are as
pronounced in childhood as later in life.
21
References
Antecol, H., & K. Bedard. (2006). Unhealthy assimilation: why do immigrants converge to
American health status levels? Demography, 43, 337–360.
Banks, J., Marmot, M.G., Oldfield, Z., & J.P. Smith. (2006) Disease and disadvantage in the
United States and England. JAMA, 295, 2037–2045.
Bleakley, H., & A. Chin. (2010). Age at arrival, English proficiency, and social assimilation
among U.S. immigrants. American Economic Journal: Applied Economics, 2, 165–192.
Burns, D.M., Lee, L., Shen, L.S., Gilpin, E., Tolley, H.D., Vaughn, J., & T.G. Shanks. (1997).
Cigarette smoking behavior in the United States. In Monograph 8: Changes in cigaretterelated disease risks and their implications for prevention and control, edited by David
M. Burns, Lawrence Garfinkel, and Jonathan M. Samet. Bethesda, MD: National Cancer
Institute, Chapter 2.
Ceballos, M., & A. Palloni. (2010). Maternal and infant health of Mexican immigrants in the
USA: the effects of acculturation, duration, and selective return migration. Ethnicity and
Health, 15, 377–396.
Cheung, B.Y., Chudek, M., & S.J. Heine. (2011). Evidence for a sensitive period for
acculturation: younger immigrants report acculturating at a faster rate. Psychological
Science, 22, 147–152.
Chiswick, B.R., Lee, Y.L., & P.W. Miller. (2006). Immigrant selection systems and immigrant
health. Contemporary Economic Policy, 26, 555–578.
Cho, Y., Frisbie, W.P., Hummer, R.A., & R.G. Rogers. (2004). Nativity, duration of residence,
and the health of Hispanic adults in the United States. International Migration Review,
38, 184–211.
22
Cho, Y., & R.A. Hummer. (2001). Disability status differentials across fifteen Asian and Pacific
Islander groups and the effect of nativity/duration. Social Biology, 48, 171–195.
Congressional Research Service (2011). The changing demographic profile of the United States.
Available at: http://www.fas.org/sgp/crs/misc/RL32701.pdf [Accessed Aug. 11, 2012]
David, R.J., & J.W. Collins. (1997). Differing birth weight among infants of U.S.-born blacks,
African-born blacks, and U.S.-born whites. New England Journal of Medicine, 337,
1209–1214.
Elders, M.J., Perry, C.L., Eriksen, M.P., & G.A. Giovino. (1994). The report of the Surgeon
General: preventing tobacco use among young people. American Journal of Public
Health, 84, 543–547.
Glick, J.E., Bates, L., & S.T. Yabiku. (2009). Mother's age at arrival in the United States and
early cognitive development. Early Childhood Research Quarterly, 24, 367–380.
Goel, M.S., McCarthy, M.P., Phillips, R.S., & C.C. Wee. (2004). Obesity among U.S. immigrant
subgroups by duration of residence. Journal of the American Medical Association, 292,
2860–2867.
Hamilton, T.G., & R.A. Hummer. (2011). Immigration and the health of U.S. black adults: does
country of origin matter? Social Science and Medicine, 73, 1551–1560.
Jasso, G., Massey, D., Rosenzweig, M., & J. Smith. (2004). Immigrant health: selectivity and
acculturation, in Anderson, N., Bulatao, R., and Cohen, B. (Eds.) Critical perspectives on
racial and ethnic differences in health in late life (pp. 227–266). Washington, D.C.: The
National Academy of Sciences.
Joseph, K.S., Demissie, K., & M.S. Kramer. (2002). Obstetric intervention, stillbirth, and
preterm birth. Seminars in Perinatology, 26(4), 250–259.
23
Julian, R., & G. Easthope. (1996). “Migrant health” in C. Grbich, (Ed.), Health in Australia (pp.
103–125). Sydney: Prentice Hall.
Kandel, D.B., L.A. Warner, & R.C. Kessler. (1998). The epidemiology of substance use and
dependence among women. In Wetherington, C.L. and Roman, A.B., (Eds.) Drug
Addiction Research and the Health of Women. Rockville, MD: National Institute of Drug
Abuse Research Monograph.
Kaushal, N. (2009). Adversities of acculturation? Prevalence of obesity among immigrants.
Health Economics, 18, 291–303.
Landale, N.S., Oropesa, R.S., & B.K. Gorman. (1999). Immigration and infant health: birth
outcomes of immigrant and native-born women, in Hernandez, D.J. (ed.) Children of
immigrants: health, adjustment and public assistance. Washington, DC: National
Academy Press. pp 244–285.
Landale, N.S., Oropesa, R.S., Llanes, D. & B.K. Gorman. (1999). Does Americanization have
adverse effects on health? stress, health habits, and infant health outcomes among Puerto
Ricans. Social Forces, 78, 613–641.
Landale, N.S., R.S. Oropesa, & B.K. Gorman. (2000). Migration and infant death: assimilation
or selective migration among Puerto Ricans? American Sociological Review, 65, 888–
909.
Lara, M., Gamboa, C., Kahramanian, M.I., Morales, L.S., & D.E.H. Bautista. (2005).
Acculturation and Latino health in the United States: a review of the literature and its
sociopolitical context. Annual Review of Public Health, 26, 367–397.
Livingston, G., & D. Cohn. (2010). The new demography of American motherhood. Pew
Research Center: A Social and Demographic Trends Report. Available at:
24
http://pewresearch.org/pubs/1586/changing-demographic-characteristics-americanmothers [Accessed Aug. 11, 2012]
Martin, J.A.,Hamilton, B.E., & M. J.K. Osterman. (2012). Three decades of twin births in the
United States, 1980–2009. NCHS Data Brief, No. 80. Available at:
http://www.cdc.gov/nchs/data/databriefs/db80.pdf [Accessed Aug. 11, 2012]
Martin, J.S., Hamilton, B.E., Sutton, P.D., Ventura, S.J., Mathews, T.J., & M.J.K. Osterman.
(2010). Births: final data for 2008. National Vital Statistics Reports 59(1). Available at:
http://www.cdc.gov/nchs/data/nvsr/nvsr59/nvsr59_01.pdf [Accessed Aug. 11, 2012].
Martinson, M., Teitler, J., & N. E. Reichman. (2011). Health across the life span in the United
States and England. American Journal of Epidemiology, 173(8), 858–865.
McCormick M.C., & J. Brooks-Gunn. (1999). Concurrent child health status and maternal recall
of events in infancy. Pediatrics, 104, 1176–1181.
McDonald, T.M., & S. Kennedy. (2004). Insights into the ‘healthy immigrant effect’: health
status and health service use of immigrants to Canada. Social Science & Medicine, 59,
1613–1627.
National Research Council. (2010). International Differences in Mortality at Older Ages:
Dimensions and Sources. E.M. Crimmins, S.H. Preston, and B. Cohen, Eds. Panel on
Understanding Divergent Trends in Longevity in High-Income Countries. Committee on
Population, Division of Behavioral and Social Sciences and Education. Washington, DC:
The National Academies Press.
Nazroo, J., & S. Karlsen. (2001). Ethnic inequalities in health: social class, racism, and identity.
Research Findings of the Economic & Social Research Council, 10, September 2001.
25
Newbold, K.B. (2005). Self-rated health within the Canadian immigrant population: risk and the
healthy immigrant effect. Social Science & Medicine, 60, 1359–1370.
O’Sullivan, J.J., Pearce, M.S., & L. Parker. (2000). Parental recall of birth weight: how accurate
is it? Archives of Diseases of Children, 82, 202–203.
Reichman, N.E. (2005). Low birth weight and school readiness. The Future of Children, 15, 91–
116.
Reichman, N.E., & D. Pagnini. (1997). Maternal age and birth outcomes: data from New Jersey.
Family Planning Perspectives, 29, 268–272 & 295.
Reijneveld, S.A., & A.H. Schene. (1998). Higher prevalence of mental disorders in
socioeconomically deprived areas of the Netherlands: community or personal
disadvantage? Journal of Epidemiology and Community Health, 52, 2–7.
Roshania, R., Narayan, K.M., & R. Oza-Frank. (2008). Age at arrival and risk of obesity among
US immigrants. Obesity, 16, 2669–2675.
Singh, G.K., & M. Siahpush (2002). Ethnic-immigrant differentials in health behaviors,
morbidity, and cause-specific mortality in the United States: an analysis of two national
data bases. Human Biology, 74, 83–109.
Singh, G.K., & S.M. Yu. (1996). Adverse pregnancy outcomes: differences between US- and
foreign-born women in major US racial and ethnic groups. American Journal of Public
Health, 86, 837–843.
Teitler, J., Hutto, N., & N.E. Reichman. (2012). Birthweight of children of immigrants by
maternal duration of residence in the United States. Social Science and Medicine, 75,
459–468.
26
United Nations. (2011). World Population Prospects: The 2010 Revision. Data available at:
http://esa.un.org/wpp/unpp/panel_indicators.htm.
Uretsky, M., & S. Mathiesen. (2007). The effects of years lived in the United States on the
general health status of California’s foreign-born populations. Journal of Immigrant and
Minority Health, 9, 125–136.
World Bank. (1999). Curbing the Epidemic: Governments and the Economics of Tobacco
Control. Washington, D.C.: World Bank.
27
Table 1. Sample Sizes of Immigrant Mothers
Giving Birth in the U.S., NHIS 1998–2009,
by Arrival Cohort
N
Arrival cohort
(rounded to nearest 100)
1955-1959
1960-1964
1965–1969
1970–1975
1975–1979
1980–1984
1985–1989
1990–1994
1995–1999
2000–2004
2005–2009
200
400
800
1500
2300
3300
4200
4200
3000
1600
200
28
Figure 1. Percent Low Birthweight in the United States, 1950–2009
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
1950
1952
1954
1956
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
0
Source: Vital Statistics of the United States, various years. Available at:
http://www.cdc.gov/nchs/products/vsus.htm
Note: Missing data covering the late 1960s and early 1970s were interpolated
29
Figure 2. Proportion of Legal Immigrants to U.S. by Region of Origin and Year of Arrival
100%
90%
80%
Oceania
70%
Africa
60%
South America
Central America
50%
Caribbean
40%
Mexico
30%
Canada
Asia
20%
Europe
10%
2010
2007
2004
2001
1998
1995
1992
1989
1986
1983
1980
1977
1974
1971
1968
1965
1962
1959
1956
1953
1950
0%
Sources: 1950: Annual Report of the INS, 1951, Table 13. 1951–1956: Annual Report of the INS, 1956,
Table 4. 1957–1959: Annual Report of the INS, 1959, Table 13. 1960–1971: Annual Report of the INS,
Table 6A. 1972–1977: INS Public Use Tapes. 1978: Statistical Yearbook of the INS, Table 6A. 1979: INS
Public Use Tapes. 1980: Statistical Yearbook of the INS, 1981, Table 2. 1981–1983: Statistical Yearbook of
the INS, 1985, Table IMM1. 1984: Statistical Yearbook of the INS, 1984, Table IMM2.4. 1985: Statistical
Yearbook of the INS, 1985, Table IMM2.4. 1986–1988: INS Public Use Tapes. 1990–1991: Statistical
Yearbook of the INS, Table 8. 1991–1997: INS Statistical Yearbook, Table 9.
30
Figure 3. Legal Immigrant Inflows to the U.S. by Region of Origin and Year of Arrival
2,000,000
1,800,000
1,600,000
Oceania
1,400,000
Africa
1,200,000
South America
Central America
1,000,000
Caribbean
800,000
Mexico
600,000
Canada
Asia
400,000
Europe
200,000
2010
2007
2004
2001
1998
1995
1992
1989
1986
1983
1980
1977
1974
1971
1968
1965
1962
1959
1956
1953
1950
0
Sources: See notes to Figure 2.
31
Figure 4. Regional Composition of Immigrant Arrival Cohorts over Time, NHIS Sample of Immigrant
Women Who Gave Birth in the U.S. between 1980 and 2009
100%
90%
80%
70%
Africa
60%
Latin & S. America
50%
Mexico
Asia
40%
Mideast
30%
Europe & Russia
20%
10%
0%
1965
1970
1975
1980
1985
1990
1995
2000
2005
Source: Authors’ calculations
32
Figure 5. Percent with High School Education by Arrival Cohort, NHIS Sample of Immigrant Women
Who Gave Birth in the U.S. between 1980 and 2009
100
90
80
70
60
50
40
30
20
10
0
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
Source: Authors’ calculations
33
Figure 6. Low Birthweight by Immigrant Cohort Within Duration Intervals, All Immigrant Mothers
12
percent low birthweight
10
8
Years in US
at time of
birth
6
0 to 2
3 to 5
4
6 to 10
11 to 15
2
16+
0
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
Year of arrival
34
Figure 7a. Estimated Effects of Duration of U.S. Residence on Low Birthweight Among All Immigrant
Mothers, Controlling for Year of Arrival and Maternal Age at Time of Birth
0
Logg Odds (reference group = 1 year)
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Years in U.S.
Note: Derived from estimates in column (a) of Appendix Table 1.
35
Figure 7b. Estimated Effects of Duration of U.S. Residence on Low Birthweight Among Hispanic
Mothers, Controlling for Year of Arrival and Maternal Age at Time of Birth
0.1
Logg Odds (reference group = 1 year)
0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Years in U.S.
Notes: Note: Derived from estimates in column (b) of Appendix Table 1. As explained in the text,
Hispanics include mothers from Mexico, Central and South America, or the Caribbean.
36
Figure 7c. Estimated Effects of Duration of U.S. Residence on Low Birthweight Among Asians,
Controlling for Year of Arrival and Maternal Age at Time of Birth
Logg Odds (reference group = 1 year)
0
-0.5
-1
-1.5
-2
-2.5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Years in U.S.
Notes: Derived from estimates in column (c) of Appendix Table 1. As explained in the text, Asians include
mothers from China, Southeast Asia, or the Indian subcontinent.
37
Figure 7d. Estimated Effects of Duration of U.S. Residence on Low Birthweight Among Mothers Who
Arrived at Ages < 18 Years, Controlling for Year of Arrival and Maternal Age at Time of Birth
1
Logg Odds (reference group = 1 year)
0.8
0.6
0.4
0.2
0
-0.2
-0.4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Years in U.S.
Note: Derived from estimates in column (d) of Appendix Table 1.
38
Figure 7e. Estimated Effects of Duration of U.S. Residence on Low Birthweight Among Mothers Who
Arrived at Ages 18+ Years, Controlling for Year of Arrival and Maternal Age at Time of Birth
0
Logg Odds (reference group = 1 year)
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.7
-0.8
-0.9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Years in U.S.
Note: Derived from estimates in column (e) of Appendix Table 1.
39
Appendix Table 1. Multivariate Estimates of the Effects of Duration of U.S. Residence, Year of Arrival, and Maternal Age on
Low Birthweight
All
Hispanics
Asians
Arrived <18
Arrived 18+
(a)
(b)
(c)
(d)
(e)
Years in U.S. (1 omitted)
2
-0.27**
-0.14
-0.74*
-0.25
-0.26*
(0.14)
(0.17)
(0.41)
(0.52)
(0.15)
3
-0.28**
-0.15
-0.14
0.46
-0.33**
(0.14)
(0.18)
(0.36)
(0.48)
(0.15)
4
-0.31**
-0.03
-1.08**
0.02
-0.29*
(0.15)
(0.18)
(0.45)
(0.51)
(0.16)
5
-0.23
-0.07
-0.52
0.56
-0.26
(0.15)
(0.20)
(0.42)
(0.52)
(0.16)
6
-0.24
0
-0.68
0.29
-0.22
(0.16)
(0.20)
(0.43)
(0.48)
(0.17)
7
-0.57***
-0.33
-1.59**
0.05
-0.54***
(0.18)
(0.22)
(0.66)
(0.53)
(0.20)
8
-0.22
-0.20
-0.52
0.49
-0.24
(0.17)
(0.22)
(0.46)
(0.51)
(0.19)
9
-0.37**
0.02
-1.63***
0.26
-0.35*
(0.17)
(0.21)
(0.62)
(0.52)
(0.20)
10
-0.31*
-0.11
-1.52**
0.51
-0.35
(0.18)
(0.22)
(0.59)
(0.51)
(0.21)
11
-0.49**
-0.53**
-0.65
0.21
-0.47*
(0.20)
(0.24)
(0.50)
(0.51)
(0.25)
12
-0.40**
-0.30
-0.80
0.20
0.32
(0.19)
(0.25)
(0.50)
(0.53)
(0.23)
13
-0.37*
-0.36
-1.00*
0.83*
-0.83***
(0.19)
(0.25)
(0.57)
(0.50)
(0.27)
14
-0.17
-0.06
-0.59
0.69
-0.25
(0.21)
(0.24)
(0.60)
(0.54)
(0.27)
15
-0.27
-0.32
-2.35**
0.61
-0.4
(0.24)
(0.29)
(1.06)
(0.53)
(0.34)
16+
-0.13
-0.11
-0.68
0.69
-0.22
(0.15)
(0.19)
(0.44)
(0.49)
(0.23)
Year of Arrival (1955-60 omitted)
1960-65
0.13
(0.50)
1965-70
0.43
(0.47)
1970-75
0.41
(0.47)
1975-79
0.54
(0.46)
1980-84
0.68
(0.46)
1985-89
0.66
(0.46)
1990-94
0.71
(0.47)
1995-99
0.81*
(0.47)
2000-04
0.85*
(0.47)
2005-09
0.72
(0.55)
-0.05
(0.60)
0.08
(0.55)
0.13
(0.53)
-0.16
(0.52)
0.30
(0.52)
0.20
(0.52)
0.10
(0.52)
0.24
(0.53)
0.18
(0.54)
0.37
(0.64)
-1.94
(1.49)
-1.41
(1.23)
-1.47
(1.16)
-0.87
(1.12)
-1.16
(1.13)
-0.95
(1.13)
-0.86
(1.14)
-1.05
(1.15)
-0.91
(1.16)
-1.17
(1.33)
0.14
(0.50)
0.45
(0.49)
0.41
(0.49)
0.6
(0.48)
0.68
(0.49)
0.52
(0.50)
0.76
(0.51)
1.07**
(0.53)
0.53
(0.67)
1.73*
(0.94)
--0.58
(0.66)
-0.22
(0.42)
-0.27
(0.36)
0.01
(0.33)
0.04
(0.33)
0.04
(0.32)
0.10
(0.32)
0.20
(0.33)
--
Mother’s age at time of birth (20-34 omitted)
< 20
0.12
-0.06
1.29*
0.36*
0.13
(0.13)
(.15)
(0.45)
(0.17)
(0.36)
35+
0.33***
0.40*
0.10
0.36*
0.35***
(0.08)
(0.10)
(0.21)
(0.17)
(0.09)
Constant
-2.91***
-2.46***
-0.78
-3.75***
-2.21***
(0.47)
(0.54)
(1.18)
(0.68)
(0.33)
N (rounded)
20800
12600
3000
7700
12100
Notes: Figures are logit coefficients and standard errors that were used to derive Figures 7a -e. *** p<0.01, ** p<0.05, * p<0.10. As
explained in the text, Hispanics include mothers from Mexico, Central and South America, or the Caribbean and Asians include
mothers from Asia, Southeast Asia, or the Indian subcontinent .
40